RedPajama-Data: An Open Source Recipe to Reproduce LLaMA training dataset

This repo contains a reproducible data receipe for the RedPajama data, with the following token counts:

Dataset	Token Count
Commoncrawl	878 Billion
C4	175 Billion
GitHub	59 Billion
Books	26 Billion
ArXiv	28 Billion
Wikipedia	24 Billion
StackExchange	20 Billion
Total	1.2 Trillion

Data Preparation

In data_prep, we provide all pre-processing scripts and guidelines.

Tokenization

In tokenization, we provide an example of how to tokenize the dataset using the GPT-NeoX tokenizer.

Visualization

In viz, we provide a dashboard for exploring a subset of the data using Meerkat.

License

The code in this repo is licensed under the Apache 2.0 license. Unless otherwise noted,

Copyright 2023 Together Computer, ETH Zürich, Stanford University

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

   http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

The file data_prep/book/dedup.py was co-developed with Ontocord.ai.

Copyright 2023 Ontocord.ai, Together Computer, ETH Zürich, Stanford University

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

   http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

The dataset itself, please refer to the licenses of the data subsets you use.

Common Crawl Foundation Terms of Use
C4 license
GitHub was limited to MIT, BSD, or Apache licenses only
Books: the_pile_books3 license and pg19 license
ArXiv Terms of Use
Wikipedia License
StackExchange license on the Internet Archive

For full terms, see the LICENSE file. If you have any questions, comments, or concerns about licensing please contact us.

To cite RedPajama, please use:

@software{together2023redpajama,
  author = {Together Computer},
  title = {RedPajama: An Open Source Recipe to Reproduce LLaMA training dataset},
  month = April,
  year = 2023,
  url = {https://github.com/togethercomputer/RedPajama-Data}
}

Acknowledgement

We are appreciative to the work done by the growing open-source AI community that made this project possible. That includes: